PMT Construction

7/29/2019 PMT Construction

1/12

INTRODUCTIONAmong photosensitive devices in use today, the photomultipliertube (or PMT) is a versatile device providing ultra-fast responseand extremely high sensitivity. A typical photomultiplier tube con-sists of a photoemissive cathode (photocathode) followed by fo-cusing electrodes, an electron multiplier and an electron collec-tor (anode) in a vacuum tube, as shown in Figure 1.When light enters the photocathode, the photocathode emitsphotoelectrons into the vacuum. These photoelectrons are thendirected by the focusing electrode voltages towards the electronmultiplier where electrons are multiplied by a secondary emis-sion process. The multiplied electrons then are collected by theanode as an output signal.Because of secondary-emission multiplication, photomultipliertubes provide extremely high sensitivity and exceptionally lownoise compared to other photosensitive devices currently usedto detect radiant energy in the ultraviolet, visible, and near infra-red regions. The photomultiplier tube also features fast time re-sponse and a choice of large photosensitive areas.This section describes the prime features of photomultiplier tubeconstruction and basic operating characteristics.

ELECTRON MULTIPLIERThe superior sensitivity (high current amplification and high S/Nratio) of photomultiplier tubes is due to the use of a low-noiseelectron multiplier which amplifies electrons by a cascade sec-ondary emission process. The electron multiplier consists of 8 to19 stages of electrodes called dynodes.There are several principal types in use today.

1) Circular-cage typeThe circular cage is generally used for the side-on type ofphotomultiplier tube. The prime features of the circular-cageare compactness, fast response and high gain obtained at arelatively low supply voltage.

2) Box-and-grid typeThis type consists of a train of quarter cylindrical dynodesand is widely used in head-on type photomultiplier tubes be-cause of good electron collection efficiency and excellent uni-formity.

CONSTRUCTIONThe photomultiplier tube generally has a photocathode in eithera side-on or a head-on configuration. The side-on type receivesincident light through the side of the glass bulb, while the head-on type receives light through the end of the glass bulb. In gen-eral, the side-on type photomultiplier tube is relatively low pricedand widely used for spectrophotometers and general photomet-

ric systems. Most side-on types employ an opaque photoca-thode (reflection-mode photocathode) and a circular-cage struc-ture electron multiplier (see description of "ELECTRON MULTI-PLIER") which has good sensitivity and high amplification at arelatively low supply voltage.The head-on type (or the end-on type) has a semitransparentphotocathode (transmission-mode photocathode) depositedupon the inner surface of the entrance window. The head-ontype provides better uniformity (see page 9) than the side-ontype having a reflection-mode photocathode. Other features ofhead-on types include a choice of photosensitive areas rangingfrom tens to hundreds of square centimeters.Variants of the head-on type having a large-diameter hemisphe-rical window have been developed for high energy physics ex-periments where good angular light reception is important.

TPMHC0006EA

TPMOC0077EB

TPMOC0078EA

a) Reflection Mode

b) Transmission Mode

TPMSC0029EA

TPMHC0084EB

Side-On Type Head-On Type

3) Linear-focused typeThe linear-focused type features extremely fast responsetime and is widely used in applications where time resolutionand pulse linearity are important. This type also has the ad-vantage of providing a large output current.

TPMOC0079EA

Figure 1: Cross-Section of Head-On Type PMT

Figure 3: Types of Photocathode

FOCUSING ELECTRODE

LAST DYNODE STEM PIN

STEMANODEELECTORON MULTIPLIER

(DYNODES)

PHOTOCATHODE

FACEPLATE

DIRECTIONOF LIGHT

SECONDARYELECTRON

VACUUM(10 Pa)

e-

-4

Figure 2: External Appearance

REFLECTION MODEPHOTOCATHODE

DIRECTION OF LIGHT

PHOTOELECTRON

SEMITRANSPARENTPHOTOCATHODE

DIRECTIONOF LIGHT

PHOTOELECTRON

4 TPMSF0039 TPMHF0192

a) Side-On Type b) Head-On Type

PHOTO-HOTO-SENSITIVEENSITIVEAREAREA

PHOTO-HOTO-SENSITIVEENSITIVEAREAREA

Construction and Operating Characteristics

PHOTO-SENSITIVE

AREAPHOTO-SENSITIVE

AREA


2/12

Hybrid dynodes combining two of the above dynodes are alsoavailable. These hybrid dynodes combine the best features ofeach dynode type.

SPECTRAL RESPONSEThe photocathode of a photomultiplier tube converts energyfrom incident light into electrons. The conversion efficiency (pho-tocathode sensitivity) varies with the wavelength of the incidentlight. This relationship between photocathode sensitivity and wa-velength is called the spectral response characteristic. Figure 4shows the typical spectral response of a bialkali photomultipliertube. The spectral response on long wavelengths is determinedby the photocathode material and on short wavelengths by thewindow material. Typical spectral response characteristics forvarious types of photomultiplier tubes are shown on pages 114and 115. In this catalog, the long-wavelength cutoff of the spec-tral response characteristic is defined as the wavelength at

which the cathode radiant sensitivity is 1 % of the maximum sen-sitivity in bialkali and Ag-O-Cs photocathodes, and 0.1 % of themaximum sensitivity in multialkali photocathodes.Spectral response characteristics shown at the end of this cata-log are typical curves for representative tube types. Actual datamay be different from tube to tube.

4) Venetian blind typeThe venetian blind type has a large dynode area and is pri-marily used for tubes with large photocathode areas. It offersbetter uniformity and a larger output current. This structure isusually used when time response is not a prime considera-tion.

5) Mesh typeThe mesh type has a structure of fine mesh electrodesstacked in close proximity. There are two mesh types of dy-node: a coarse mesh type and a fine mesh type. Both typesprovide improved pulse linearity and high resistance to mag-netic fields. The mesh type also has position-sensitive capa-bility when used with cross-wire anodes or multiple anodes.The fine mesh type is particularly suited for use in applica-

tions where high magnetic fields are present.

6) Microchannel plate (MCP) (see page 58)The MCP is a thin disk consisting of millions of microglasstubes (channels) fused in parallel with each other. Eachchannel acts as an independent electron multiplier. The MCPoffers much faster time response than other discrete dy-nodes. It also features good immunity from magnetic fieldsand two-dimensional detection ability when multiple anodesare used.

TPMOC0080EA

7) Metal Channel typeThe metal channel dynode has a compact dynode construc-tion manufactured by our unique fine machining techniques.It delivers high-speed response due to a space between eachdynode stage that is much smaller than other types of con-ventional dynodes. The metal channel dynode is also idealfor position sensitive measurement.

TPMOC0084EA

TPMOC0081EB

TPMOB0070EA

TPMOC0082EA

FINE-MESH TYPE

ELECTRONELECTRON

ELECTRON

COARSE MESH TYPE

1mm 13m

ELECTRON

Figure 4: Typical Spectral Response of Bialkali Photocathode

2000.01

0.1

1

10

100

400 600 00

(HEAD-ON TYPE, BIALKALI PHOTOCATHODE)

WAVELENGTH (nm)

CATHODERADIANTSENSITIV

ITY(mA/W)

QUANTUMEFFICIENCY

(%)

CATHODERADIANT

QUANTUMEFFICIENCY

5


3/12

PHOTOCATHODE MATERIALSThe photocathode is a photoemissive surface usually consistingof alkali metals with very low work functions. The photocathodematerials most commonly used in photomultiplier tubes are asfollows:

1) Ag-O-CsThe transmission-mode photocathode using this material isdesignated S-1 and sensitive from the range of visible light to

infrared radiation (300 mm to 1000 nm). The reflection modecovers a slightly narrower range from 300 mm to 1100 nm.Since Ag-O-Cs has comparatively high thermionic dark emis-sion (refer to "ANODE DARK CURRENT" on page 8), pho-tomultiplier tubes of this photocathode material are chieflyused for detection in the infrared region with the photo-cathode cooled.

2) GaAsGaAs activated in cesium is also used as a photocathode.The spectral response of this photocathode material usuallycovers a wider spectral response range than multialkali, fromultraviolet to 930 nm, which is comparatively flat over therange between 300 mm and 850 nm.

3) InGaAsThis photocathode material has greater extended sensitivityin the infrared range than GaAs. Moreover, in the range bet-

ween 900 mm and 1000 nm, InGaAs has a much higher S/Nratio than Ag-O-Cs.

4) Sb-CsSb-Cs has a spectral response in the ultraviolet to visiblerange and is mainly used in reflection-mode photocathodes.

5) Bialkali (Sb-Rb-Cs, Sb-K-Cs)These materials have a spectral response range similar tothe Sb-Cs photocathode, but have higher sensitivity and low-er dark current than Sb-Cs. They also have a blue sensitivityindex matching the scintillation flashes of NaI scintillators,and so are frequently used for radiation measurement usingscintillation counting.

6) High temperature bialkali or low noise bialkali (Na-K-Sb)This is particularly useful at higher operating temperaturessince it can withstand up to 175 C. One major application isin the oil well logging industry. At room temperatures, thisphotocathode operates with very low dark current, making itideal for use in photon counting applications.

7) Multialkali (Na-K-Sb-Cs)The multialkali photocathode has a high, wide spectralresponse from the ultraviolet to near infrared region. It is wide-ly used for broad-band spectrophotometers and photon count-ing applications. The long wavelength response can beextended to 930 nm by special photocathode activation pro-cessing.

8) Cs-Te, Cs-IThese materials are sensitive to vacuum UV and UV rays butnot to visible light and are therefore referred to as solar blind.Cs-Te is quite insensitive to wavelengths longer than 320 nm,and Cs-I to those longer than 200 nm.

WINDOW MATERIALSWindow materials commonly used in photomultiplier tubes aredescribed below. The window material must carefully be select-ed according to the application because the window materialdetermines the spectral response short wavelength cutoff.

RADIANT SENSITIVITY AND QUANTUM EFFICIENCYAs Figure 4 shows, spectral response is usually expressed interms of radiant sensitivity or quantum efficiency as a function ofwavelength. Radiant sensitivity is the photoelectric current fromthe photocathode, divided by the incident radiant power at agiven wavelength, expressed in A/W (amperes per watt). Quan-tum efficiency (QE) is the number of photoelectrons emitted fromthe photocathode divided by the number of incident photons.Quantum efficiency is usually expressed as a percent. Quantumefficiency and radiant sensitivity have the following relationshipat a given wavelength.

LUMINOUS SENSITIVITYSince measuring the spectral response characteristic of pho-tomultiplier tubes requires a sophisticated system and a greatdeal of time, we instead provide figures for anode or cathodeluminous sensitivity and only provide spectral response charac-teristics when specially required by the customer.Cathode luminous sensitivity is the photoelectric current from thephotocathode per incident light flux (10-5 to 10-2 lumens) from atungsten filament lamp operated at a distribution temperature of2856K. Anode luminous sensitivity is the anode output current(amplified by the secondary emission process) per incident lightflux (10-10 to 10-5 lumens) on the photocathode. Although thesame tungsten lamp is used, the light flux and the applied vol-tage are adjusted to an appropriate level. These parameters areparticularly useful when comparing tubes having the same or

where S is the radiant sensitivity in A/W at the given wavelength

and is the wavelength in nm (nanometers).

1) Borosilicate glassThis is the most frequently used window material. Borosi-licate glass transmits radiation from the infrared to approx-imately 300 nm. It is not suitable for detection in the ultravio-let region. For some applications, a combination of a bialkaliphotocathode and a low-noise borosilicate glass (so calledK-free glass) is used. The K-free glass contains very lowpotassium (40K) which can cause unwanted backgroundcounts. Tubes designed for scintillation counting oftenemploy K-free glass not only for the faceplate but also for theside bulb to minimize noise pulses.

2) UV-transmitting glass (UV glass)This glass as the name implies is ideal for transmitting ultra-violet radiation and is used as widely as a borosilicate glass.The UV cutoff is approximately 185 nm.

3) Synthetic silicaThe synthetic silica transmits ultraviolet radiation down to 160nm and offers lower absorption in the ultraviolet range com-pared to fused silica. Since the synthetic silica has a differentthermal expansion coefficient than Kovar, which is used for

the tube leads, it is not suitable as the tube stem material(see Figure 1 on page 4). Borosilicate glass is used for thestem, and a graded seal using glass with gradually differentthermal expansion coefficients is connected to the syntheticsilica window. The graded seal structure is vulnerable toshock so the tube should be handled carefully.

4) MgF2 (magnesium fluoride)Crystals of alkali halide are superior in transmitting ultravioletradiation, but have the disadvantage of deliquescence.Among these crystals, MgF2 is known as a practical windowmaterial because it offers low deliquescence and transmitsultraviolet radiation down to 115 nm.

TPMOB0076EB

QE= 100S 1240

Figure 5: Typical Transmittance of Various Window Materials

MgF2

UV-TRANSMITTING

GLASS

BOROSILICATEGLASS

100 120 160 200 240 300 400 500

100

10

1

TRANSMITTANCE(%)

WAVELENGTH (nm)

SYNTHETICQUARTZ

6



4/12

BLUE SENSITIVITY INDEX AND RED/WHITE RATIOThe cathode blue sensitivity index and the red/white ratio areoften used as a simple comparison of photomultiplier tube spec-tral response.The cathode blue sensitivity index is the photoelectric currentfrom the photocathode produced by a light flux of a tungstenlamp at 2856K passing through a blue filter (Corning CS 5-58 pol-ished to half stock thickness), measured under the same condi-

tions as the cathode luminous sensitivity measurement. The lightflux, once transmitted through the blue filter cannot beexpressed in lumens. The blue sensitivity index is an importantparameter in scintillation counting using an NaI scintillator sincethe NaI scintillator produces emissions in the blue region of thespectrum, and may be the decisive factor in energy resolution.The red/white ratio is used for photomultiplier tubes with a spec-tral response extending to the near infrared region. This parame-ter is defined as the quotient of the cathode sensitivity measuredwith a light flux of a tungsten lamp at 2856K passing through ared filter (Toshiba IR-D80A for the S-1 photocathode or R-68 forothers) divided by the cathode luminous sensitivity measuredwithout filters under the same conditions as in cathode luminoussensitivity measurement.

GAIN (CURRENT AMPLIFICATION)Photoelectrons emitted from a photocathode are accelerated byan electric field so as to strike the first dynode and produce sec-ondary electron emissions. These secondary electrons thenimpinge upon the next dynode to produce additional secondaryelectron emissions. Repeating this process over successivedynode stages achieves a high current amplification. A verysmall photoelectric current from the photocathode can thereforebe observed as a large output current from the anode of the pho-tomultiplier tube.Gain is simply the ratio of the anode output current to the photo-electric current from the photocathode. Ideally, the gain of a pho-tomultiplier tube having n dynode stages and an average second-ary emission ratio per stage is n. While the secondary electronemission ratio is given by =AE

where A is the constant, E is the interstage voltage, and is acoefficient determined by the dynode material and geometricstructure. This usually has a value of 0.7 to 0.8.When a voltage V is applied between the cathode and the anodeof a photomultiplier tube having n dynode stages, the gain ,becomes

Since photomultiplier tubes generally have 9 to 12 dynode stag-es, the anode output has a 6th to 10th power gain proportional tothe input voltage. So just a slight fluctuation in the applied vol-tage will appear as magnified 6 to 10 times in the photomultipliertube output. This means the photomultiplier tube is extremelysusceptible to fluctuations in the power supply voltage, so thepower supply must be extremely stable and provide a minimumripple, drift and temperature coefficient. Various types of well-regulated high-voltage power supplies designed for theserequirements are available from Hamamatsu (see page 96).

similar spectral response range. Hamamatsu final test sheetsaccompanying the tubes usually indicate these parametersexcept for tubes with Cs-I or Cs-Te photocathodes insensitive totungsten lamp light. (Radiant sensitivity at a specific wavelengthis listed for those tubes using CsI or Cs-Te.)The cathode luminous sensitivity is expressed in uA/lm (microam-peres per lumen) and anode luminous sensitivity is expressed inA/lm (amperes per lumen). Note that the lumen is a unit used forluminous flux in the visible region and therefore these values

may be meaningless for tubes that are sensitive beyond the vis-ible light region.

TPMOB0054EC

TPMOB0055EB

TPMOB0058EB

(K: constant)

Figure 6: Typical Human Eye Responseand Spectral Distribution of 2856K Tungsten Lamp

100

80

60

40

0

20

200 400 600 800 1000 1200 1400

WAVELENGTH (nm)

RELATIVEVALUE(%)

VISUAL SENSITIVITY

TUNGSTENLAMPAT 2856 K

Figure 7: Transmittance of Various Filters

Figure 8: Typical Gain vs. Supply Voltage

WAVELENGTH (nm)

TRANSMITTANCE(%)

100

80

60

40

20

0200 400 600 800 1000 1200

CORNINGCS-5-58(1/2 STOCKTHICKNESS)

TOSHIBA R-68

TOSHIBAIR-D80A

= n = (A E)n = A ( ) nV

n+1

= Vn = K VnAn

(n+1)n

200 300 500 700 1000 1500

104

103

102

101

100

10-1

10-2

109

108

107

106

105

104

103

ANODELUMINOUSSENSITIVITY

(A/lm)

SUPPLY VOLTAGE (V)

GAIN

GAIN

ANODE LUMINOUSSENSITIVITY

7


5/12

ANODE DARK CURRENTA small amount of current flows in a photomultiplier tube evenwhen the tube is operated in a completely dark state. This outputcurrent is called the anode dark current, and the resulting noiseis a critical factor in determining the lower limit of light detection.As Figure 9 shows, dark current is greatly dependent on the sup-ply voltage.

Major sources of dark current may be categorized as follows:

The anode dark current decreases with time after the tube isplaced in a dark state. In this catalog, anode dark currents aremeasured after 30 minutes of storage in a dark state.

For tubes listed in this catalog, the value of ENI may be calculat-ed by the above equation. Usually it has a value between 10-15and 10-16 watts (at the peak sensitivity wavelength).

ENI (EQUIVALENT NOISE INPUT)ENI indicates the photon-limited signal-to-noise ratio. ENI refersto the amount of light in watts necessary to produce a signal-to-noise ratio of unity in the output of a photomultiplier tube. The val-ue of ENI is given by:

MAGNETIC FIELD EFFECTSMost photomultiplier tubes are affected by the presence of mag-netic fields. Magnetic fields may deflect electrons from their nor-mal trajectories and cause a loss of gain. The extent of the gainloss depends on the type of photomultiplier tube and its orienta-tion in the magnetic field. Figure 11 shows typical effects of mag-netic fields on some types of photomultiplier tubes. In general,tubes having a long path from the photocathode to the firstdynode (such as large diameter tubes) tend to be more adverse-ly affected by magnetic fields.

1) Thermionic emission of electronsThe materials of the photocathode emit tiny quantities of ther-mionic electrons even at room temperature. Most dark cur-rents originate from the thermionic emissions, especiallythose from the photocathode since they are successively mul-tiplied by the dynodes. Cooling the photocathode is mosteffective in reducing thermionic emission and is particularlyuseful in applications where low dark current is essentialsuch as in photon counting.Figure 10 shows the relationship between dark current andtemperature for various photocathodes. Photocathodes whichhave high sensitivity in the red to infrared region, especiallyS-1, show higher dark current at room temperature. Photomul-tiplier tubes using these photocathodes are usually cooledduring operation.Hamamatsu provides thermoelectric coolers (C659 series,C4877, C4878) designed for various sizes of photomultipliertubes (see page 102).

TPMOB0071EB

TPMOB0065EB

2) Ionization of residual gases (ion feedback)Residual gases inside a photomultiplier tube can be ionizedby collision with electrons. When these ions strike the photo-cathode or earlier stages of dynodes, secondary electronsmay be emitted. These secondary electrons result in relative-ly large output noise pulses. These noise pulses are usuallyobserved as afterpulses following the primary signal pulsesand may be a problem in detecting short light pulses. Presentphotomultiplier tubes are designed to minimize afterpulses.

3) Glass scintillationWhen electrons deviating from their normal trajectories strikethe glass envelope, scintillations may occur and a dark pulsemay result. To eliminate this type of dark pulse, photomultipli-er tubes may be operated with the anode at a high voltageand the cathode at ground potential. But this is not alwayspossible during tube operation. To obtain the same effectwithout difficulty, Hamamatsu developed an "HA coating" inwhich the glass bulb is coated with a conductive paint makingthe same electrical potential as the cathode (see "GROUNDPOLARITY AND HA COATING" on page 11).

4) Leakage current (ohmic leakage)Leakage current resulting from imperfect insulation of theglass stem base and socket may be another source of darkcurrent. This is predominant when the photomultiplier tube is

operated at a low voltage or low temperature. The flatterslopes in Figure 9 and 10 are mainly due to leakage current.Contamination from dirt and moisture on the surface of thetube stem, base or socket may increase the leakage current,and should therefore be avoided.

5) Field emissionsWhen a photomultiplier tube is operated at a voltage near themaximum rated value, electrons might be emitted from elec-trodes by the strong electric field and cause dark pulses. Sooperating the photomultiplier tube at a voltage 20 to 30% low-er than the maximum rating is recommended.

electronic charge (1.60 10-19 coul.)anode dark current in amperes after 30minutestorage in darknessgainbandwidth of the system in hertz (usually 1 hertz)anode radiant sensitivity in amperes per wattat the wavelength of interest

where q =Idb =

=f=S =

ENI = (watts)2q Idb f

S

Figure 9: Typical Dark Current vs. Supply Voltage

Figure 10: Anode Dark Current vs. Temperature

APPLIED VOLTAGE (V)

ANODEDARKCURRENT(nA)

(AFTER 30 MINUTE STORAGE)

40010-3

600 800 1000 1200 1400

10-2

10-1

100

101

TEMPERATURE (C)

ANODEDARKCURRENT(A)

-40-60 -20 0 20 40

R3550A(HEAD-ON TYPE,LOW-NOISE BIALKALI)

R316-02(HEAD-ON TYPE, Ag-O-Cs)

R6095(HEAD-ON TYPE, BIALKALI)

R374(HEAD-ON TYPE,MULTIALKALI)

10-5

10-7

10-6

10-8

10-9

10-10

10-12

10-11

10-13

8



6/12

When a photomultiplier tube has to be operated in magneticfields, it may be necessary to shield the tube with a magneticshield case. (Hamamatsu provides a variety of magnetic shieldcases. See page 106). The magnetic shielding factor is used toexpress the effect of a magnetic shield case. This is the ratio ofthe strength of the magnetic field outside the shield case orHout, to that inside the shield case or Hin. The magnetic shield-ing factor is determined by the permeability , the thickness t(mm) and inner diameter r (mm) of the shield case as follows.

Note that the magnetic shielding effect decreases towards theedge of the shield case as shown in Figure 12. Covering thetube with a shield case longer than the tube length by at leasthalf the tube diameter is recommended.

Hamamatsu provides photomultiplier tubes using fine-mesh typedynodes (see page 54). These photomultiplier tubes exhibitmuch higher resistance to external magnetic fields than the pho-tomultiplier tubes with other dynodes. When the light level to bemeasured is high, "triode" and "tetrode" type tubes can be usedeven in highly magnetic fields.

SPATIAL UNIFORMITYAlthough the focusing electrodes of a photomultiplier tube aredesigned so that electrons emitted from the photocathode ordynodes are collected efficiently by the first or followingdynodes, some electrons may deviate from their desired trajecto-ries causing lower collection efficiency. The collection efficiencyvaries with the position on the photocathode from which the pho-toelectrons are emitted and influences the spatial uniformity of aphotomultiplier tube. The spatial uniformity is also determined bythe photocathode surface uniformity itself.In general, head-on type photomultiplier tubes provide better spa-tial uniformity than side-on types because of the photocathode tofirst dynode geometry. Tubes especially designed for gammacamera applications have excellent spatial uniformity, becauseuniformity is the decisive factor in the overall performance of agamma camera.

TEMPERATURE CHARACTERISTICSDark current originating from thermionic emissions can be

reduced by decreasing the ambient temperature of a photomult ip-lier tube. The photomultiplier tube sensitivity also varies with thetemperature, but these changes are smaller than temperature-induced changes in dark current, so cooling a photomultipliertube will significantly improve the S/N ratio.In the ultraviolet to visible region, the sensitivity temperature coef-ficient has a negative value, while near the long wavelength cut-off it has a positive value. Figure 14 shows typical temperaturecoefficients for various photocathodes versus wavelength, meas-ured at room temperatures. Since the change in temperaturecoefficient change is large near the long wavelength cutoff, tem-perature control may be needed in some applications.

TPMOB0086EC

TPMHC0085EB TPMSC0030EC

TPMOB0011EB

TPMOB0013EB

HoutHin

3 t4 r

=

Figure 11: Typical Effects by Magnetic Fields Perpendicularto Tube Axis

Figure 13: Examples of Spatial Uniformity

Figure 13: Examples of Spatial Uniformity

Figure 12: Edge Effect of Magnetic Shield Case

120

110

100

90

80

70

60

50

40

30

20

10

0-0.5 -0.4 -0.3 -0.2 -0.1 0.1 0.2 0.3 0.40 0.5

( )

( )

28 mm dia.SIDE - ON TYPE

LINEAR-FOCUSEDTYPE DYNODE

13 mm dia.HEAD-ON TYPE

CIRCULAR CAGETYPE DYNODE

38 mm dia.HEAD-ON TYPE

MAGNETIC FLUX DENSITY (mT)

RELATIVEOUTPUT

(%)

LONGER than r

EDGE EFFECT

1000

100

10

1

t

L

r r

SHIELDINGFACTOR(Ho/Hi)

2r PHOTOMULTIPLIER TUBE

1) Head-On Type(R6231-01 for gamma camera)

2) Side-On TypeReflection-mode photocathode

100

50

0

ANODESENSITIVITY(%)

PHOTO-CATHODE(TOP VIEW)

ANO

DE

SEN

SITIVITY(%)

50ANODESENSITIVITY (%

PHOTO-CATHODE

GUIDE KEY

500 1000

100

1.5

1

0.5

0

-0.5

-1200 300 400 500 600 700 800 900 1100 12001000

Cs-Te

Sb-Cs

BIALKALIMULTIALKALI

GaAs (Cs)

Ag-O-Cs

WAVELENGTH [nm]

TEMPERATURECOEFFICIENT

FORANODESENSITIVITY[%

C]

Sb-CsMULTIALKALI

9


7/12

HYSTERESISPhotomultiplier tubes exhibit a slightly unstable output for sever-al seconds to nearly 1 minute after a voltage is applied or light isinput, and the output may overshoot or undershoot before reach-ing a stable level (Figure 15). This unstable condition is calledhysteresis and may be a problem in spectrophotometry andother applications.Hysteresis is mainly caused by electrons deviating from theirplanned trajectories and electrostatically charging the dynode

support section and glass bulb. When the applied voltage chan-ges along with a change in the input light, noticeable hysteresiscan occur.As a countermeasure, many Hamamatsu side-on photomultipliertubes employ an "anti-hysteresis design" which virtually elimin-ates hysteresis.

TIME RESPONSEIn the measurement of pulsed light, the anode output signalshould faithfully reproduce a waveform resembling the incidentpulse waveform. This reproducibility is greatly affected by theelectron transit time, anode pulse rise time, and electron transittime spread (TTS).As illustrated in Figure 17, the electron transit time is the time in-terval between the arrival of a delta function light pulse (pulsewidth less than 50 ps) at the photocathode and the instant when

the anode output pulse reaches its peak amplitude. The anodepulse rise time is defined as the time needed to rise from 10 %to 90 % of peak amplitude when the entire photocathode is illu-minated by a delta function light pulse (pulse width less than 50ps).The electron transit time fluctuates between individual light pul-ses. This fluctuation is called transit time spread (TTS) and de-fined as the FWHM of the frequency distribution of electron tran-sit times (Figure 18). The TTS is an important factor in time-re-solved measurement.The time response characteristics depend on the dynode struc-ture and applied voltage. In general, photomultiplier tubes usinga linear-focused or circular-cage structure exhibit better time re-sponse than tubes using a box-and-grid or venetian blind struc-ture. MCP-PMTs, which employ an MCP in place of conventional

dynodes, offer better time response than tubes using other dy-nodes. For example, these have a significantly better TTS com-pared to normal photomultiplier tubes because a nearly parallelelectric field is applied between the photocathode, the MCP andthe anode. Figure 19 shows typical time response characteristicsvs. applied voltage for Hamamatsu R2059 (51 mm diameterhead-on, 12-stage, linear-focused type).

DRIFT AND LIFE CHARACTERISTICWhile operating a photomultiplier tube continuously over a longperiod, the anode output current of the photomultiplier tube mayvary slightly over time, even though operating conditions havenot changed. Among the anode current fluctuations, changesover a relatively short time are called "drift", while changes overlong periods such as 103 to 104 hours or more are called the lifecharacteristic. Figure 16 shows typical drift curves.

Drift is primarily caused by damage to the last dynode by heavyelectron bombardment. Therefore the use of lower anode currentis desirable. When stability is of prime importance, keeping theaverage anode current within 1 A or less is recommended.

TPMOC0071EA

TPMHB0448EB

TPMOB0060EB

TPMHB0126EC

IiI min.

I max.

0 5 6 7

ANODECURRENT

TIME (MINUTE)

TIME (h)

RELATIVE

ANODECURRENT(%)

1 10 100

100

0

01000 10000

x +

x -

x

RISE TIME FALL TIME

ANODEOUTPUT

SIGNAL90 %

10 %

TRANSIT TIME

DELTA FUNCTION LIGHT

10


Figure 15: Hysteresis

Figure 17: Anode Pulse Rise Time and Electron Transit Time

Figure 18: Electron Transit Time Spread (TTS)Figure 16: Typical Life Characteristics

-5

100

101

102

103

104

5-4 -3 -2 -1 0 1 2 3 4

TIME (ns)

RELAT

IVECOUNT

TYPE NO. : R2059 550 ps 1228 ps


8/12

VOLTAGE-DIVIDER CIRCUITSInterstage voltages for the dynodes of a photomultiplier tube areusually supplied by voltage-divider circuits consisting of series-

connected resistors. Schematic diagrams of typical voltage-div-ider circuits are illustrated in Figure 20. Circuit (a) is a basic ar-rangement (DC output) and (b) is for pulse operations. Figure 21shows the relation between the incident light level and the outputcurrent of a photomultiplier tube using the voltage-divider circuit(a). Deviation from ideal linearity occurs at a certain incident lev-el (region B). This is caused by an increase in dynode voltagedue to the redistribution of the voltage loss between the last fewstages, resulting in an apparent increase in sensitivity. As the in-put light level is increased, the anode output current begins tosaturate near the value of the current flowing through the voltagedivider (region C). To prevent this problem, it is recommendedthat the voltage-divider current be maintained at least at 20times the average anode output current required from the photo-multiplier tube.

GROUND POLARITY AND HA COATINGThe general technique used for voltage-divider circuits is toground the anode with a high negative voltage applied to thecathode, as shown in Figure 20. This scheme facilitates the con-nection of such circuits as ammeters or current-to-voltage con-version operational amplifiers to the photomultiplier tube. How-ever, when a grounded anode configuration is used, bringing agrounded metallic holder or magnetic shield case near the bulbof the tube can cause electrons to strike the inner bulb wall, re-sulting in the generation of noise. Also, in head-on type photo-multiplier tubes, if the faceplate or bulb near the photocathode isgrounded, the slight conductivity of the glass material causes acurrent to flow between the photocathode (which has a highnegative potential) and ground. This may cause significant dete-rioration of the photocathode. For this reason, extreme care isrequired when designing housings for photomultiplier tubes andwhen using electrostatic or magnetic shield cases..In addition, when using foam rubber or similar material to mountthe tube in its housing, it is essential that material having suffi-ciently good insulation properties be used. This problem can besolved by applying a black conductive coat around the bulb, con-necting it to the cathode potential and covering the bulb with aprotective film. This is called an "HA Coating" (see Figure 23).

Generally high output current is required in pulsed light applica-tions. In order to maintain dynode potentials at a constant valueduring pulse durations and obtain high peak currents, capacitorsare placed in parallel with the divider resistors as shown in Fig-ure 20 (b). The capacitor values depend on the output charge.When the output linearity versus input pulsed light needs to bebetter than 1 %, the capacitor value should be at least 100 timesthe photomultiplier output charge per pulse. If the peak outputcurrent (amperes) is I, the pulse width (seconds) t, and the vol-tage across the capacitor (volts) V, then the capacitor value Cshould be as follows:

In high energy physics applications where a high pulse output isrequired, output saturation will occur at a certain level as the in-cident light is increased while the interstage voltage is keptfixed,. This is caused by an increase in electron density betweenthe electrodes, causing space charge effects which disturb theelectron current flow. As a corrective measure to overcomethese space charge effects, the voltage applied to the last fewstages, where the electron density becomes high, should be setto a higher value than the standard voltage distribution so that

the voltage gradient between those electrodes is enhanced. Forthis purpose, a so-called tapered divider circuit (Figure 22) is of-ten employed. Use of this tapered divider circuit improves pulselinearity 5 to 10 times better than in normal divider circuits.Hamamatsu provides a variety of socket assemblies incorporat-ing voltage-divider circuits. They are compact, rugged, light-weight and carefully engineered to obtain the maximum perfor-mance of a photomultiplier tube with just a simple connection.

TPMOB0059EB

TACCC0035EB

b) For pulse operation

TACCC0030EC

TACCB0005EA

I tV

C > 100 (farads)

500 1000 1500 2000 30002500

0

1

210

10

10

SUPPLY VOLTAGE (V)

TIME(ns

)

T. T. S

RISE TIME

TYPE NO. : R2059

a) Basic arrangement for DC operation

-HV

1R 1R 1R 1R 1R 1R 1R 1R 1R 1R 1R

RL

RL

-HV

1R 1R 1R 1R 1R 1R 1R 1R 1R 1R 1R

C1 C2 C3

ANODEPHOTOCATHODE

ANODEPHOTOCATHODE

0.001

0.01

.1

1.0

10

0.001 0.01 0.1 1.0 10

LIGHT FLUX (A.U.)

RATIOOFAVERAGEOUTPUTCURRENT

TODIVIDERCURRENT

IDEALCURVE

C

ACTUALCURVE

11

Figure 19: Time Response Characteristics vs. Supply Voltage

Figure 20: Schematic Diagrams of Voltage-Divider Circuits

Figure 22: Typical Tapered Divider Circuit

Figure 21: Output Characteristics of PMT Using Voltage-

Divider Circuit (a)

1R

C1 C2 C3

SIGNALOUTPUT

1R 1R 1R 2R 3R 2.5R

RL

-HV

ANODEPHOTOCATHODE


9/12

As mentioned above, the HA coating can be effectively used toeliminate the effects of external potential on the side of the bulb.However, if a grounded object is located on the photocathodefaceplate, there are no effective countermeasures. Glass scintil-lation, if occurring in the faceplate, has adverse noise effectsand also causes deterioration of the photocathode sensitivity. Tosolve these problems, it is recommended that the photomultipliertube be operated in the cathode grounding scheme, as shown inFigure 24, with the anode at a high positive voltage. For exam-

ple in scintillation counting, since the grounded scintillator is di-rectly coupled to the faceplate of a photomultiplier tube, ground-ing the cathode and maintaining the anode at a high positive vol-tage is recommended. In this case, a coupling capacitor Cc mustbe used to isolate the high positive voltage applied to the anodefrom the signal, and DC signals cannot be output.

Simply counting the photomultiplier tube output pulses will not

result in an accurate measurement, since the output containsnoise pulses such as dark pulses and cosmic ray pulses extra-neous to the signal pulses representing photoelectrons asshown in Figure 27. The most effective method for eliminatingthe noise is to discriminate the output pulses according to their

A typical pulse height distribution (PHD) for a photomultipliertube output is shown in Figure 28. In this PHD, the lower leveldiscrimination (LLD) is set at the valley trough and the upper lev-el discrimination (ULD) at the foot where there are very few out-put pulses. Most pulses smaller than the LLD are noise and pul-ses larger than the ULD result from cosmic rays, etc. Therefore,by counting the pulses remaining between the LLD and ULD, ac-curate light measurements can be made. In the PHD, Hm is themean height of the pulses. The LLD should be set at 1/3 of Hmand the ULD at triple Hm. The ULD may be omitted in most ca-ses.

Considering the above, a clearly defined peak and valley in thePHD is a very significant characteristic required of photomultipli-er tubes for photon counting. Figure 28 shows the typical PHD ofa photomultiplier tube selected for photon counting.

TPMOC0015EA

TACCC0036EC

TPMOC0073EA

TPMOC0074EA

TPMOC0075EC

TPMOC0076EA

PHOTON COUNTING

Photon counting is one effective way to use a photomultipliertube for measuring extremely low light levels and is widely usedin astronomical photometry and for making chemiluminescenceand bioluminescence measurements. In its usual application, anumber of photons enter the photomultiplier tube and create anoutput pulse train like that in (a) of Figure 25. The actual outputobtained by the measurement circuit is a DC current with a fluc-tuation as shown at (b).

SCINTILLATION COUNTINGScintillation counting is one of the most sensitive and effectivemethods for detecting radiation. It uses a photomultiplier tubecoupled to a scintillator that produces light when struck by radia-tion.

When the light intensity becomes so low that the incident pho-tons are separated as shown in Figure 26. This condition iscalled a single photon event. The number of output pulses is indirect proportion to the amount of incident light and this pulsecounting method has the advantages of better S/N ratio and sta-bility than the current measurement method that averages all thepulses. This pulse counting technique is known as the photoncounting method.

GLASS BULB

CONDUCTIVE PAINT(SAME POTENTIALAS CATHODE)

INSULATINGPROTECTIVE COVER

CONNECTED TOCATHODE PIN

R1 R2 R3 R4 R5 R7

RP

+HV

Cc

C

CR6

SIGNALOUTPUT

ANODEPHOTOCATHODE

TIME

LLD

ULD

ULD: Upper Level Discri.LLD: Lower Level Discri.

SIGNAL PULSE

COSMIC RAY PULSE

PULSEHEIGHT

TIME

DARK CURRENTPULSE

12


Figure 23: HA Coating

Figure 26: Discrete Output Pulses (Single Photon Event)

Figure 27: Output Pulse and Discrimination Level

Figure 28: Typical Single Photon Pulse Height Distribution

Figure 24: Cathode Ground Scheme

Figure 25: Overlapping Output Pulses

b)

a)

TIME

TIME

SIGNAL PULSE + NOISE PULSE

NOISE PULSE

LLD Hm ULD

COUNTS

PULSE HEIGHT


10/12

In radiation particle measurements, there are two parametersthat should be measured. One is the energy of individual radia-tion particles and the other is the amount of radiation. Radiationmeasurement should determine these two parameters.When radiation particles enter the scintillator, they produce lightflashes in response to each particle. The amount of flash is ex-tremely low, but is proportional to the energy of the incident par-ticle. Since individual light flashes are detected by thephotomultiplier tube, the output pulses obtained from the photo-multiplier tube contain information on both the energy andamount of pulses, as shown in Figure 30. By analyzing these

output pulses using a multichannel analyzer (MCA), a pulseheight distribution (PHD) or energy spectrum is obtained, andthe amount of incident particles at various energy levels can bemeasured accurately. Figure 31 shows typical PHDs or energyspectra when radiation (55Fe, 137Cs, 60Co) is detected by thecombination of an NaI(Tl) scintillator and a photomultiplier tube.The PHD must show distinct peaks at each energy level. Thesepeaks are evaluated as pulse height resolution which is the mostsignificant characteristic in the radiation measurements. As Fig-ure 32 shows, the pulse height resolution is defined as theFWHM (a) divided by the peak value (b) when pulse height dis-tribution is measured using a single radiation source such as137Cs and 55Fe.

Pulse height resolution is mainly determined by the quantum effi-ciency of the photomultiplier tube that detects the scintillatoremission. In the case of thallium-activated sodium iodide or Na-I(Tl), which is one of the most popular scintillators, a head-ontype photomultiplier tube with a bialkali photocathode is widelyused since its spectral response matches the NaI(Tl) scintillatorspectrum.

TPMHC0052EB

TPMOB0087EC

TPMOB0088EB

TPMOB0073EA

TPMOC0039EC

a) 55Fe+NaI (TI)

b) 137Cs+NaI (Tl)

c) 60Co+NaI (Tl)

Energy resolution = 100 %ab

SCINTILLATOR

PMT

THE HEIGHT OF OUTPUTPULSE IS PROPORTIONALTO THE ENERGY OFINCIDENT PARTICLE.

TIMECU

RRENT

TIME

REFLECTIVECOATING

PHOTOCATHODE

PHOTOELECTRONS

DYNODESANODE

PMT

GAMMA RAY

RADIATIONSOURCE

SCINTILLATOR

OPTICAL COUPLING(USING SILICONE OIL etc.)

1000

500

mm a. 51 mm t)

500 1000

ENERGY

COUNTS

10000

5000

(51 mm dia. 51 mm t)

5000 1000ENERGY

COUNTS

10000

5000

500

ENERGY

COUNTS

1000

(51 mm dia. 51 mm t)

PULSE HEIGHT

NUMBEROFPULSES

b

aH

H

2

0.1

WAVELENGTH (nm)

Q

UANTUM

EFFICIENCY(%)

RELAT

IVEEMISSIONDISTRIBUTION

OFVARIOUSSCINTILLATOR(%)

1

10

100

200 300 400 500 600 700 800

BaF

BIALKALI

CsI (Tl)

BGO

13

Figure 29: Scintillation Detector Using PMT and Scintillator

Figure 32: Definition of Pulse Height Resolution

Figure 33: PMT Spectral Response and Spectral Emission of

Scintillators

Figure 30: Incident Radiation Particles and PMT Output

Figure 31: Typical Pulse Height Distributions (Energy Spectra)


11/12

LOAD RESISTANCE

Since the output of a photomultiplier tube is a current signal andthe type of external circuit to which photomultiplier tubes areusually connected has voltage inputs, a load resistor is used forcurrent-voltage conversion. This section describes factors toconsider when selecting this load resistor.Since for low output current levels, the photomultiplier may beassumed to act as virtually an ideal constant-current source, theload resistance can be made arbitrarily large, when converting alow-level current output to a high-level voltage output. In prac-tice, however, using a very large load resistance causes poorfrequency response and output linearity as described below.

HIGH-SPEED OUTPUT CIRCUITS

When detecting high-speed and pulsed light signals, a coaxialcable is used to make the connection between the photomultipli-er tube and the electronic circuit. Since commonly used cableshave characteristic impedances of 50 or 75 , this cable mustbe terminated in a pure resistance equal to the characteristic im-pedance to match the impedance and ensure distortion-freetransmission of the signal waveform. If a matched transmissionline is used, the impedance of the cable as seen by the photo-multiplier tube output will be the characteristic impedance of thecable, regardless of the actual cable length so no distortion willoccur in the signal waveform.If the impedance is not properly matched when the signal is re-ceived, the impedance seen at the photomultiplier tube outputwill differ depending on both frequency and cable length, caus-ing significant waveform distortion. Impedance mismatchesmight also be due to the connectors being used. So these con-nectors should be chosen according to the frequency range tobe used, to provide a good match with the coaxial cable.When a mismatch at the signal receiving end occurs, not all ofthe pulse energy from the photomultiplier tube is dissipated atthe receiving end and is instead partially reflected back to thephotomultiplier tube via the cable. However if an impedancematch has been achieved at the cable end on the photomultipliertube side, then this reflected energy will be fully dissipated there.. If this is a mismatch, however, the energy will be reflected andreturned to the signal-receiving end because the photomultipliertube itself acts as an open circuit. Since part of the pulse makesa round trip in the coaxial cable and is again input to the receiv-ing end, this reflected signal is delayed with respect to the mainpulse and results in waveform distortion (so called ringing phe-nomenon).

To prevent this phenomenon, in addition to matching the impe-dance at the receiving end, a resistor is needed for matching thecable impedance at the photomultiplier tube end as well (Figure36). If this is provided, it is possible to eliminate virtually all ring-ing caused by an impedance mismatch, although the outputpulse height of the photomultiplier tube is reduced to one-half ofthe normal level by use of this impedance matching resistor.

This value of Ro, which is less than the value of RL, is then theeffective load resistance of the photomultiplier tube. If, for exam-ple, RL=Rin, then the effective load resistance is 1/2 that of RLalone. From this we see that the upper limit of the load resis-tance is actually the input resistance of the amplifier and thatmaking the load resistance much greater than this value doesnot have a significant effect.While the above description assumed the load and input impe-dances to be purely resistive, stray capacitances, input capaci-tance and stray inductances affect the phase relationships dur-ing actual operation. Therefore, as the frequency is increased,these circuit elements must be considered as compound impe-dances rather than pure resistances.From the above, three guides can be derived for selecting theload resistance:

1) When frequency response is important, the load resistanceshould be made as small as possible.

2) When output linearity is important, the load resistance shouldbe chosen to keep the output voltage within a few volts.

3) The load resistance should be less than the input impe-dance of the external amplifier.

This relationship indicates that even if the photomultiplier tubeand amplifier have very fast response, the response will belimited to the cutoff frequency fc of the output circuit. If the loadresistance is made large, then the voltage drop across R Lbecomes large at high current levels, affecting the voltagedifferential between the last dynode stage and the anode. Thisincreases the effect of the space charge and lowers the

efficiency of the anode in collecting electrons. In effect, theoutput becomes saturated above a certain current, causing pooroutput linearity (output current linearity versus incident lightlevel) especially when the circuit is operated at low voltages.

In Figure 35, let us consider the effect of the internal resistanceof the amplifier. If the load resistance is RL and the inputimpedance of the amplifier is Rin, the combined parallel outputresistance of the photomultiplier tube, Ro, is given by thefollowing equation.

In the circuit of Figure 34, if we let the load resistance be R L andthe total capacitance of the photomultiplier tube anode to allother electrodes including stray capacitance such as wiringcapacitance be Cs, then the cutoff frequency fc is expressed bythe following relationship.

TACCC0037EB

TACCC0017EA

TACCC0039EB

1)

2)

12 Cs RL

fc =

RL RinRL + Rin

Ro =

-HV

SIGNALOUTPUT

Ip LR CS

ANODEPHOTOCATHODE

PMT P

DYn

RL CSRin

SIGNALOUTPUT

SIGNALOUTPUT

RinCSRL

PMT P

DYnCC

PMT

HOUSING ANTI-REFLECTIONRESISTOR

50 OR 75 CONNECTOR

50 OR 75 COAXIAL CABLE

RL

(50 OR 75

MATCHING RESISTOR)

14

Connections to External Circuits

Figure 34: Photomultiplier Tube Output Circuit

Figure 35: Amplifier Internal Resistance

Figure 36: Connection to Prevent Ringing


12/12

If the operational amplifier has an offset current (Ios), the above-described output voltage becomes Vo = -R f (Ip+Ios), with the off-set current component being superimposed on the output. Fur-thermore, the magnitude of the temperature drift may create aproblem. In general, a metallic film resistor which has a low tem-perature coefficient is used for the resistance R f, and for high re-sistance values, a vacuum-sealed type with a low leakage cur-rent is used. Carbon resistors with their highly temperature-de-pendent resistance characteristics are not suitable for this appli-cation.In addition to the above factors, when measuring extremely lowlevel currents such as 100 pA and below, the materials used tofabricate the circuit also require careful selection. For example,

materials such as bakelite are not suitable. More suitable materi-als include teflon, polystyrol or steatite. Low-noise cables shouldalso be used, since general-purpose coaxial cables exhibit noisedue to physical factors. An FET input operational amplifier is rec-ommended for measuring low-level current.

In Figure 39, if a capacitance Cf (including any stray capacitance)is in parallel with the resistance Rf, the circuit exhibits a timeconstant of (Rf Cf), and the response speed is limited to thistime constant. This is a particular problem if the Rf is large. Straycapacitance can be reduced by passing Rf through a hole in ashield plate. When using coaxial signal input cables, oscillationsmay occur and noise might be amplified since the cable capaci-tance Cc and Rf are in a feedback loop. While one method to

avoid this is to connect C f in parallel with Rf, to reduce high fre-quency gain as described above, this method creates a timeconstant of Rf Cf which limits the response speed.

OPERATIONAL AMPLIFIERS

When a high-sensitivity ammeter is not available, using an op-erational amplifier allows making measurements with an inex-pensive voltmeter. This section explains the technique for con-verting the output current of a photomultiplier tube to a voltagesignal. The basic circuit is as shown in Figure 38, for which theoutput voltage, Vo, is given by the following relationship.

Vo = -Rf Ip

This relationship is derived for the following reason. If the inputimpedance of the operational amplifier is extremely large, andthe output current of the photomultiplier tube is allowed to flowinto the inverted () input terminal of the amplifier, most of thecurrent will flow through Rf and subsequently to the operationalamplifier output circuit. The output voltage Vo is therefore givenby the expression -Rf Ip. When using such an operational am-plifier, it is not of course possible to make unlimited increases inthe output voltage because the actual maximum output is rough-ly equal to the operational amplifier supply voltage. At the otherend of the scale, for extremely small currents, there are limitsdue to the operational amplifier offset current (Ios), the quality ofRf, and other factors such as the insulation materials used.

Next, let us consider waveform observation of high-speed pulsesusing an oscilloscope. This type of operation requires a low loadresistance. However, the oscilloscope sensitivity is limited so anamplifier may be required.Cables with a matching resistor have the advantage that thecable length will not affect the electrical characteristics of thecable. However, since the matching resistance is very low com-pared to the usual load resistance, the output voltage becomestoo small. While this situation can be remedied with a high gainamplifier, the inherent noise of such an amplifier can itself hurtmeasurement performance. In such cases, the photomultipliertube should be brought as close as possible to the amplifier toreduce stray capacitance and a larger load resistance should beused (while still maintaining the frequency response), to achievethe desired input voltage. (See Figure 37.)

It is relatively simple to implement a high-speed amplifier using awide-band video amplifier or operational amplifier. However, asa trade-off for design convenience, these ICs tend to create per-formance problems (such as noise). This makes it necessary toknow their performance limits and take corrective action if nec-essary.As the pulse repetition frequency increases, baseline shift be-comes one reason for concern. This occurs because the DC sig-nal component has been eliminated from the signal circuit bycoupling with a capacitor which blocks the DC components. Ifthis occurs, the reference zero level observed at the last stage isnot the actual zero level. Instead, the apparent zero level is atime-average of the positive and negative fluctuations of the sig-nal waveform. This is known as baseline shift. Since the height

of the pulses above this baseline level is affected by the repeti-tion frequency, this phenomenon can be a problem when ob-serving waveforms or discriminating pulse levels.

TACCC0026EA

TACCC0041EA

TACCC0042EA

PMT

PDYn

RL

WIRING SHOULD BEAS SHORT AS POSSIBLE.

OSCILLOSCOPE

PMTOP-AMP

Vo= -lp Rf

Rf

p lp lp

V+

SIGNALOUTPUT

OP-AMP.

SHIELD CIRCUIT

Rf

Cs

Cf

-

+

15

Figure 37: Measurement with Ringing Suppression Measures

Figure 38: Current-Voltage Conversion Using Operational

Amplifier

Figure 39: Frequency Compensation by Operational Amplifier

PMT Construction

Documents

Transcript of PMT Construction